Preoperative MRI versus intraoperative frozen-section in the assessment of myometrial invasion in endometrioid type endometrial cancer

In this study, it is aimed to compare the diagnostic accuracy of preoperative magnetic resonance imaging (MRI) and intraoperative frozen section in determining the depth of myometrial invasion in patients with endometrioid type endometrial cancer. Retrospective evaluation of 69 patients with endometrioid type endometrial cancer who underwent preoperative MRI and subsequently intraoperative frozen-section and surgical staging between February 2004 and September 2011. Surgical staging was performed according to 2009 FIGO classification. Final histopathological examination was accepted as the standard of reference. The median age of patients was 59 years (range 41-80 years). Histopathological evaluation revealed that 41 patients had superficial myometrial invasion (< 50% of myometrium) and 28 had deep myometrial invasion (=50% of myometrium). Sensitivity, specificity, positive predictive value, negative predictive value and accuracy were 64.2%, 87.8%, 78.2%, 78.2% and 78.2% for MRI, and 89.3%, 97.6 %, 96.2 %, 93.0 % and 94.2% for frozen-section, respectively. The diagnostic accuracy of the intraoperative frozen-section was significantly higher than that of preoperative MRI according to McNemar's test (p=0.03). A statistically significant negative correlation was found between MRI accuracy and tumor grade (p=0.0002). None of the other variables was associated with incorrect prediction of MRI. The rate of incorrect prediction of myometrial invasion with frozen-section was significantly higher in nonvisualized tumors (p=0.01). A significant difference was not detected with other variables and frozen-section. Intraoperative frozen section has a better accuracy compared to MRI in the assessment of myometrial invasion in endometrioid type endometrial cancer. MRI seems suboptimal in the identification of invasion to the myometrium.